# LetAgentPay

> AI agent spending management platform — policy middleware between AI agents and payments.

LetAgentPay lets humans define spending policies (budgets, category limits, schedules) for their AI agents. Agents send purchase requests via API; the platform validates them against policies and returns approve/reject decisions in real time. Multi-channel notifications (email, Telegram, browser push) let owners approve or reject requests from anywhere.

## Quick Start

1. Sign up at https://letagentpay.com/auth/signin (magic link, no password)
2. Create an agent in the dashboard → get a Bearer token (agt_...)
3. Integrate via SDK, MCP Server, or REST API

## Integration Options

- **Python SDK**: `pip install letagentpay` — Python integration
- **TypeScript SDK**: `npm install letagentpay` — JS/TS integration
- **MCP Server**: `npx letagentpay-mcp` — Claude Desktop, Cursor, OpenClaw
- **REST API**: Direct HTTP calls for any language/framework
- **x402 Protocol**: Policy middleware for crypto-micropayments (USDC on Base)
- **LangChain**: Custom tool (BaseTool) — see examples/langchain_tool.py
- **OpenAI Agents SDK**: Function tool (@function_tool) — see examples/openai_agents.py
- **CrewAI**: CrewAI tool (@tool) — see examples/crewai_agent.py
- **Vercel AI SDK**: `npm install @letagentpay/ai` — ready-made tools for generateText/streamText
- **Google ADK**: Plain function tools — see examples/google_adk_agent.py
- **Stripe**: Policy middleware before Stripe payments — see examples/stripe_governance.py

## Links

- Full API reference: https://letagentpay.com/llms-full.txt
- Developer docs: https://letagentpay.com/developers
- OpenAPI spec: https://letagentpay.com/api/v1/openapi.json